AI, LLMs & Cybersecurity: The Good, The Bad, and The Deepfakes with Dr. Andrea Isoni

Episode Summary:

The race to be the best in artificial intelligence (AI) is getting stronger. Open-source models are now competing with closed systems, which affects how we create new technology and keep it safe. Will open AI help us make significant discoveries quickly, or could it be too risky? How will governments, businesses, and hackers use this new technology?

In this episode, Dr. Andrea Isoni, the Chief AI Officer at AI Technologies, discusses the latest advances in AI, from computer chips and deepfakes to cybersecurity threats

Timestamps:

[02:30] - The cost and performance of new AI models vs. OpenAI

[06:45] - The LLM arms race: Open vs. Closed AI models

[12:10] - The importance of optimization in pre-training and post-training AI models

[18:00] - China's AI strategy and global AI adoption challenges

[22:40] - The battle for GPU dominance: NVIDIA, AMD, and upcoming players

[29:15] - Cybersecurity and AI: Threats and solutions

[35:20] - The impact of the EU AI Act and global AI regulations

[42:00] - Deepfakes and how AI can protect against AI-powered attacks

[48:50] - The future of AI security: Blockchain, cryptographic verification, and watermarking

Key Takeaways:

• AI Optimization is Important: The next wave of large language models (LLMs) will focus on improving optimization, using synthetic data to boost efficiency.  
• China’s AI Game Plan: Even with all the geopolitical bumps, China is making serious strides in AI and chip manufacturing. 
• Cybersecurity & AI: AI-driven cyber threats like deepfakes, phishing, and AI poisoning are becoming more sophisticated, so developing AI defenses is essential now.  
• Regulation is Key: The EU AI Act and other global regulations will play a significant role in adopting AI and developing security measures.  
• The GPU Race: While NVIDIA is still the top dog, AMD and some newer brands are catching up, prioritizing cost efficiency and innovation.

Resources & Links:

• State of Artificial Intelligence Report: stateofartificialintelligence.com
•  AI Research & Insights: AI Technologies - Click on the "AI Insights" section for more reports and analysis.
• EU AI Act Overview: European Commission AI Regulations
• OECD AI Incident Index: OECD AI Policy Observatory
• Reality Defender (Deepfake Detection Startup): Reality Defender

Be sure also to follow our hosts on LinkedIn:

👉Gianna Whitver

👉Maria Velasquez

Thank you for tuning in to another episode of Breaking Through in Cybersecurity Marketing. You can join the Cybersecurity Marketing Society on our website, main LinkedIn page, or podcast LinkedIn page, and keep up with us on Twitter.

See you in the next episode!